> ## Documentation Index
> Fetch the complete documentation index at: https://docs.heybreez.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Nodes

Nodes are the building blocks of a workflow. Each node performs a specific action, sets behavior, or determines how the call progresses. Nodes connect together to form the complete call flow.

Nodes fall into four main categories.

***

## Input & Output Components

These nodes control how a workflow begins, ends, and returns structured data.

#### Start

The first node in every workflow. It cannot be deleted.\
It sets default transcriber (Hear), chat model (Think), voice (Speak), and the initial agent/path.\
It contains four connectors: Hear, Think, Speak, and Agent Path.

#### End Call

Ends the call session immediately.

#### Tool Response

Returns structured output back to the agent.\
Ends only the current branch, not the entire call.

***

## AI Components

AI components represent the underlying models used for transcription (speech-to-text), reasoning, and voice (text-to-speech).

These nodes are added when:

* Setting the workflow defaults (branching from the Start node)
* Overriding the workflow defaults (branching from an Agent node)

#### Transcriber Nodes

* **Breez Transcriber**
* **Deepgram**
* **ElevenLabs Transcriber**
* **OpenAI Transcriber**

#### Chat Model Nodes

* **OpenAI Chat Model**
* **OpenAI Realtime Model**

#### Voice Nodes

* **Breez Voice**
* **Deepgram Voice**
* **ElevenLabs Voice**
* **OpenAI Voice**

***

## Agents

Agents represent the AI entity interacting with the caller.

#### Agent Breez

The main conversational agent, feel free to rename the node to whatever you want to name your agent (especially if you have multiple agents/personas in your workflow).\
The agent can carry its own instructions and model overrides.\
It has one output: an Agentic Tool path.

#### Agentic Tool

Defines a flow that the agent can autonomously trigger based on its reasoning and configuration.\
Inside an agentic tool branch, any Logic, Action or Agent nodes can be used.

***

## Logic & Action Nodes

These nodes define routing, decision-making, external actions, and structural behavior within the workflow.

#### If

Adds conditional branching based on specified conditions.

#### A/B Test

Splits traffic across branches for experiments or load balancing.

#### Outcome

Marks outcome as success/failure and updates the total\_outcome\_score of the session (this can be viewed in the Session History table under Project > Monitoring)

#### HTTP Request

Sends a request to an external system or API.

#### Say

Speaks a predefined message to the caller.

#### Wait

Pauses for a specified duration.

#### Background Noise

Adds configurable ambient background noise for realism.

***

## Summary

These node types work together to create flexible, powerful workflows.\
Each node contributes a specific capability—whether handling AI models, controlling call flow, or performing logical and external operations.

To understand each node more in depth, navigate to its dedicated page from the left sidebar.